If you’re using Gemini without a higher-tier subscription, you’re about to have fewer options if Google changes who gets access to the model.

According to 9to5Google’s report, Google is tightening model access for both free Gemini users and subscribers to its $4.99-per-month AI Plus plan. The company’s lightest model, Flash-Lite, will be available to people without a Google AI subscription starting October 9. Those with AI Plus will still get Flash-Lite and Flash, but not Pro.

Despite this, Google AI Pro subscribers aren’t losing anything. In fact, they’re getting something new: Deep Think, a feature previously restricted to Google’s more expensive plans.

In addition, Google is preparing to add adjustable “effort” levels for its models, giving users more control over how much computing power Gemini uses.

Together, the changes indicate that Google is drawing sharper lines between Gemini’s free, entry-level, and premium offerings.

What Is Changing With Gemini?

The details come from an updated Gemini support document that Google revised this week.

For users without a Google AI subscription, the change is straightforward. Starting October 9, they will only be able to use Gemini 3.5 Flash-Lite. They will no longer have access to Gemini 3.6 Flash or Gemini 3.1 Pro.

Despite a slightly better deal, AI Plus subscribers still lose access to some models. In the $4.99-per-month plan, Flash-Lite and Flash will remain, but Pro will no longer be included.

Google says AI Plus subscribers will receive an email explaining the changes. As such, October 9 is only applicable to those without a plan. It’s not clear when the AI Plus restriction will go into effect, according to the report.

It’s worth mentioning that the version numbers can be confusing. Flash-Lite is version 3.5, Flash is version 3.6, and Pro is version 3.1, but these numbers refer to different model families. In other words, you can’t look at “3.6” versus “3.1” and assume Flash is newer or more capable than Pro.

Here’s how access now breaks down:

Google AI plan	Flash-Lite	Flash	Pro

Without a plan

Yes

No

No

AI Plus

Yes

Yes

No

AI Pro

Yes

Yes

Yes

AI Ultra

Yes

Yes

Yes

It’s pretty obvious what the basic strategy is. Become a subscriber and you’ll unlock more advanced models. That means that as of now, Pro is a feature reserved for AI Pro subscribers and above.

AI Pro Gets Deep Think

For AI Pro and AI Ultra subscribers, Google isn’t taking anything away.

Instead, AI Pro is getting Deep Think.

Deep Think is described by Google as providing maximum parallel reasoning. Previously, the feature was only available to subscribers of Google’s $99.99-per-month and $199.99-per-month plans.

That’s a pretty substantial change.

In other words, Google is bringing a feature that previously cost five or ten times as much to the mainstream paid tier for $19.99 per month.

9to5Google’s illustration shows Gemini 3.1 Pro with Deep Think enabled on AI Ultra. The publication’s earlier coverage of Gemini 3.1 Pro provides additional background on the model.

Those who actually use Gemini for complex reasoning, research, or problem-solving may find AI Pro more compelling. And that appears to be the point.

Google Is Also Adding “Effort” Levels

It’s not just model access that’s changing. It also appears that Google will offer low, medium, and high effort settings for each model by October. Basically, you can tell Gemini how much effort you want it to put into your response.

By increasing the effort setting, the model should be able to handle difficult tasks more effectively and produce more comprehensive answers. As a tradeoff, more usage allowance is consumed.

As of right now, Gemini has an extended-thinking option for solving complex problems. By proposing a three-level system, that concept would become much more explicit.

Also, it allows Gemini to be more closely integrated with Google AI Studio and Antigravity controls. Users may find this useful. After all, every prompt does not require the maximum amount of reasoning.

In most cases, burning through a higher usage allowance makes no sense if you’re asking Gemini to summarize an email, rewrite a sentence or explain something simple. If a research question, coding challenge, or reasoning task is complex, it might be worth spending more time on computing.

However, users must now consider another resource. You aren’t simply asking, “Which model should I use?” You’re also asking, “How much effort should I spend on this?”

This Builds on Google’s May Usage Changes

It is important to remember that these restrictions don’t exist in isolation. In May, Google moved Gemini app usage limits to compute-based limits. 9to5Google covered the change at the time. That shift is relevant because the new effort controls fit neatly into the same system.

Think of it as two different gates.

Your subscription is the first gate. You can use it to determine which models you can use.

The second is the amount of usage you are allowed. This determines how much you can use them.

With low, medium, and high effort settings, Google’s second gate becomes even more granular. In contrast, a more intensive response consumes more of your allowance than a quick, low-effort response.

It’s a much more sophisticated way of managing AI usage than simply saying everyone gets X number of prompts per day. It also allows Google to manage its computing costs while making the system less restrictive.

What This Means for Free Gemini Users

For free users, there isn’t much ambiguity. You’re getting less. But, if you regularly use Flash or Pro for more demanding tasks, you’ll have to move to a paid plan to keep that access.

For those who use Gemini occasionally for basic questions, brainstorming or simple writing assistance, the difference probably won’t matter much. However, it could have a much greater impact on people who have integrated Gemini into their daily routine.

This is especially true for people who use AI to conduct research, code, analyze, or solve complex problems. Even if the model is lighter, it can still be useful, but “useful” and “equivalent” are not the same.

Ultimately, Google is treating Gemini’s free version more as an introduction than as a full version.

AI Plus Is the More Interesting Case

The most interesting changes may be found in AI Plus.

With a monthly cost of $4.99, AI Plus is well below the $19.99 AI Pro subscription. Nevertheless, if AI Plus does not include Pro, one of the biggest differences between the two tiers will be even more apparent. The difference in price is $15 per month, which is easier for Google to justify since AI Pro includes all AI Plus models, plus Pro itself and now Deep Think.

In other words, Google isn’t just taking something away from AI Plus. Compared to AI Pro, it’s considerably more valuable.

From Google’s perspective, that’s a common subscription strategy. Cheap tiers get you in the door. With a premium tier, you get the features that serious users need.

Whether AI Plus subscribers will stay in the middle tier depends on how much value they receive.

AI Pro Is Becoming the Sweet Spot

The biggest winner here is arguably AI Pro. Subscribers have access to Flash-Lite, Flash, and Flash Pro. Deep Think is also included. That’s important because Deep Think was previously associated with $99.99 and $199.99 monthly plans. The $19.99 price seems much more reasonable now.

In the future, AI Pro may become the tier Google most wants people to choose if they consistently use Gemini. But for users who want the most access, Google still offers AI Ultra, but it’s becoming less of an expensive upgrade and more like a standard subscription.

The Effort Controls Could Be a Good Thing

There’s another way to look at all of this.

Giving users control over effort isn’t necessarily a bad thing. As a matter of fact, it could make Gemini more efficient. You don’t need a deep reasoning process to ask, “What’s the capital of France?” But you may need one when Gemini is analyzing a complex business problem or debugging a complex program.

With the option to choose between low, medium, and high effort, users can determine where their usage budget should be allocated.

But there’s the catch.

The more effort you put into it, the more allowance you have left. As a result, Google has effectively turned computing power into another form of currency within Gemini. As the task becomes harder, you’re likely to spend more of your allowance.

This isn’t necessarily unreasonable. AI computing isn’t free. However, it does make usage limits more important than when the main concern was how many prompts were left.

There Are Still Some Open Questions

Not everything is settled yet.

For AI Plus subscribers, the biggest question is when Pro access will disappear. According to Google, affected subscribers will receive an email, but the report does not give a specific date.

In addition, it’s not clear how much of an AI Pro user’s allowance will be consumed by Deep Think requests. There’s also a question mark regarding effort controls. As Google states in its support documentation, low, medium, and high settings will be added in October, but they may not be available to everyone immediately.

In addition, there is an important limitation to the reporting: these changes are related to the Gemini app. According to the report, Google’s other products and services will not be subject to similar restrictions.

What Gemini Users Should Do Now

If you’re a Gemini free user who relies regularly on Flash or Pro, October 9 should be on your calendar.

Before the restriction goes into effect, you may want to complete any work you need to complete using those models.

However, AI Plus subscribers have a little more uncertainty. Watch for Google’s email and check the Gemini support documentation for the latest information. As these changes roll out, Google can update the details on the support page.

But the bigger picture is more interesting than any single model restriction. Google is turning Gemini into a much better segmented product. The lightweight experience is available to free users. Users can use the lightweight experience for free. In the middle tier, AI Plus is available. With AI Pro, you get the serious tools. And the top spot is still held by AI Ultra.

With usage increasingly tied to computing power, the days of thinking of AI access simply as “how many prompts do I get?” may be numbered.

In Google’s case, it’s a way to manage expensive AI infrastructure while attracting users who need more functionality. Everyone else, however, will have one more thing to keep track of before pressing the prompt.